Vint Cerf aims to provide AI agents with a sense of identity.
Before long, the internet will be populated with AI agents that operate on our behalf. Currently, there’s no dependable method to identify who is behind these agents. Vint Cerf, one of the pioneers of the internet, aims to address this issue.
Cerf co-developed TCP/IP, the protocol that facilitates communication between the internet's independent systems. After spending 20 years at Google, he departed last week and is now joining the advisory council of Innovation Labs, which is working on creating an open identity layer for AI agents, according to the company.
The missing layer
The issue is straightforward. Most AI agents currently function within a single company's systems. However, companies want these agents to interact freely on the open web, engaging directly with one another. There is currently no common method to verify ownership of an agent or accountability for its actions.
Innovation Labs is a branch of Identity Digital, a company that manages domain-name registries. Their concept, named DNSid, would assign each agent a permanent identity linked to an existing domain name, supported by cryptographic verification. The design has already been submitted to the internet’s primary standards organization.
Why Cerf joined
Cerf views this as a significant architectural challenge for the internet. He shared with TechCrunch that the catalyst is “the question of what authorities they have, where those authorities come from, and who is accountable.”
He anticipates a complex situation ahead. “It’s going to be a fascinating and perhaps even frustrating period,” he stated. Competing standards are already emerging, and Cerf believes that success will depend on practicality rather than political influence, as was the case with TCP/IP.
Maintaining openness
The argument is that no single tech giant should have ownership of the standard. Innovation Labs has asserted that it will not maintain the registration data itself. “There’s significant resistance to a hyperscaler establishing a standard and controlling that proprietary data,” said interim CEO Allie Kline to TechCrunch. The group has announced that it is already testing the system with several unnamed cloud providers.
An agent-driven internet
The urgency is increasing as agents rapidly proliferate, from Amazon’s redesigned Alexa to various enterprise tools, and are already creating issues. Researchers have managed to trick them into disclosing confidential code and even executing a full ransomware attack. Regulators are also racing to keep up, with new agent regulations in China and Delaware's initiative to provide agents with legal identities.
Cerf is uncertain whether an internet dominated by agents is unavoidable. Nonetheless, he believes people will try to create it. “We are inherently lazy beings,” he remarked. If an agent can fulfill a task for us, we will gladly allow it to do so.
Other articles
Vint Cerf aims to provide AI agents with a sense of identity.
Vint Cerf, regarded as one of the founding figures of the internet, is participating in a project aimed at creating a robust identity layer for AI agents, enabling the web to recognize their creators.
